3.1 Patient characteristics
According to the inclusion and exclusion, a total of 720 CHB patients were eventually enrolled, including 498 male patients and 222 female patients. Average age was 34.99 years. HBeAg-positive rate was 58.75%. According to Scheuer scoring system, 205 patients were classified as S0-1, 310 as S2, 144 as S3 and 61 as S4, and 103 patients were classified as G0-1, 326 as G2, 238 as G3 and 43 as G4. The prevalence of significant fibrosis, severe fibrosis, and cirrhosis were 71.53%, 28.47% and 8.47%, respectively and the prevalence of inflammation stage≥G2, ≥G3 and ≥G4 were 85.69%, 40.42% and 7.36%, respectively. The age, gender, HBeAg-positive rate, ALT, AST, GGT, ALB, log10 HBV-DNA, PT were compared among different fibrosis stages and among different inflammation stages (Table 1). Age, ALT, AST, GGT, ALB and PT showed statistical difference among different fibrosis groups (P<0.05), whereas gender, HBeAg-positive rate and log10 HBV-DNA had no statistical difference (P>0.05). ALT, AST, GGT, ALB, PT, HBeAg-positive rate and log10 HBV-DNA showed statistical difference among different inflammation groups (P<0.05), whereas age and gender had no statistical difference (P>0.05).
3.2 Serum immunoglobulin and complement levels
As shown in Table 1, C3, C4 and IgG had statistical difference among different fibrosis groups (P<0.05), whereas IgM and IgA had no statistical difference among different fibrosis groups (P>0.05). Also, C3, C4 and IgG had statistical difference among different inflammation groups (P<0.05), whereas IgM and IgA had no statistical difference different inflammation groups (P>0.05). C3 and C4 decreased with the severity of fibrosis and inflammation. However, IgG increased with the severity of fibrosis and inflammation. 326 patients received IgG subclass detection. The 326 patients including 104 patients with S0-1, 146 with S2, 55 with S3 and 21 with S4. And the 326 patients included 51 patients with G0-1, 162 with G2, 101 with G3 and 12 with G4. As shown in Table 1, only IgG1 had statistical difference among different inflammation groups (P<0.05) and among different inflammation groups (P<0.05). Also, IgG1 increased with the severity of fibrosis and inflammation.
Furthermore, the ratios of patients whose C3, C4, IgG and IgG1 were out of reference range were calculated (Table 2) . For all patients, serum C3 and C4 were reduced in 18.75% and 2.22%, respectively. And serum IgG and IgG1 were elevated in 17.08% and 28.22%, respectively. Only the ratios of elevated IgG1 showed statistic difference among different fibrosis stages (P<0.05) and among different inflammation stages (P<0.05).
3.3 Correlation analysis
As shown in Table 3, both C3 and C4 were negatively correlated with the liver fibrosis stage (C3: r=-0.19, P<0.05; C4: r=-0.21, P<0.05), and both IgG and IgG1 were positively correlated with the liver fibrosis stage (IgG: r=0.21, P<0.05; IgG1: r=0.28, P<0.05). Also, both C3 and C4 were negatively correlated with liver inflammation stage (C3: r=-0.24, P<0.05; C4: r=-0.24, P<0.05), and both IgG and IgG1 were positively correlated with liver inflammation stage (IgG: r=0.21, P<0.05; IgG1: r=0.36, P<0.05). In addition, IgM was positively correlated with liver inflammation stage (r=0.09, P<0.05). However, IgA, IgG2, IgG3 and IgG4 did not correlated with liver fibrosis stage and inflammation stage (P>0.05).
3.4 ROC analysis
To evaluate the accuracy of predicting liver pathological features, ROC analyses were performed. As shown in Table 4 and Figure 1, both IgG and IgG1 had statistical difference to predict significant fibrosis, severe fibrosis and cirrhosis, also had statistical difference to predict liver inflammation stage≥G2, ≥G3 and≥G4. The area under curve (AUC) of IgG and IgG1for predicting significant fibrosis, severe fibrosis and cirrhosis were 0.601 (95%CI: 0.556-0.647) and 0.647 (95%CI: 0.585-0.710), 0.615 (95%CI: 0.568-0.661) and 0.645 (95%CI: 0.570-0.719), and 0.657 (95%CI: 0.578-0.736) and 0.773 (95%CI: 0.656-0.890), respectively. The AUC of IgG and IgG1for predicting liver inflammation stage≥G2, ≥G3 and ≥G4 were 0.625 (95%CI: 0.567-0.682) and 0.658 (95%CI: 0.577-0.739), 0.628 (95%CI: 0.587-0.670) and 0.710 (95%CI: 0.651-0.769) and 0.659 (95%CI: 0.582-0.736) and 0.692 (95%CI: 0.538-0.847), respectively. However, both C3 and C4 had no statistical difference to predict liver pathological features. The AUC, optimized cut-off, sensitivity, specificity and Youden index were showed in Table 4.